Better directory detection semantics and zip behavior (if we keep it)

[colearendt]

Current / legacy pins behavior:

• would try to zip a directory
• would allow the zip = TRUE argument to "force" the zip behavior

However,

• zip = TRUE is a confusing way to tell pins "I want to publish a directory, please zip it up for me"
• The implementation using withr::with_dir() in the zip flow did not work, because it changes the working directory sufficiently to break the inherent directory names / zip name / .zip output location, etc.

I had to change this (saved_model is a directory):

library(pins)
library(keras)

# Pin it on RStudio Connect
pins::board_register_rsconnect()
pins::pin(
  "saved_model",
  "peptide_model",
  "Peptide Prediction Model",
  "rsconnect",
  zip = TRUE
)

To this:

library(pins)
library(keras)
# Pin it on RStudio Connect
pins::board_register_rsconnect()
local_zipfile <- zip::zip("model.zip", "saved_model")
pins::pin(
  local_zipfile,
  "peptide_model",
  "Peptide Prediction Model",
  "rsconnect"
)

[hadley]

My thought is that each pin should be one "thing", but that thing might be a directory. I'm not sure how much work this will be; I think it's mostly testing it and fixing all the places that I assumed that a pin would be a file.

[hadley]

Superseded by #427, I think.

[github-actions[bot]]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.